Although the idea of wavelength division multiplexing has been extensively developed at lower frequencies, and has previously been suggested for optical frequencies for use with optical fibers, the prior art has not been able to solve the problems associated with producing a practical multiplexer. In particular, it was not known how to achieve a clean separation of adjacent channels (i.e., to reduce crosstalk), how to reduce the insertion loss to an acceptable level or how to make a multiplexer small enough, to be compatible with the other components of an optical fiber system.
There is prior art that discloses a multiplexed communications system that employs a prism for wavelength separation (U.S. Pat. No. 3,845,294 issued Oct. 29, 1974 to G. S. Indig et al.). There is also a pulse-shaping optical system that employs a reflection diffraction grating for wavelength separation but does not provide for multiplexing (U.S. Pat. No. 3,863,063 issued Jan. 28, 1975 to G. S. Indig et al.). It would be desirable to have a smaller device and a lower insertion loss than is available with these systems.
The closest prior art considered the use of diffraction gratings and rejected them as providing too much attenuation. "Wavelength Division Multiplexing in Light Interface Technology" by R. J. Stigliani et al in Report AD-721 085, available from the National Technical Information Service, addresses the problem of designing a multiplexer. They conclude that diffraction gratings are impractical, thus teaching away from the present invention. Their preferred method involves a combination of narrowband generators and narrowband receivers.